| 00:13 | <jsbell> | Does "... and abort these steps" add anything useful after "return" or "throw"? HTML uses "return foo and abort these steps" which I assume is just old-and-crufty style, but maybe people like it... ? |
| 00:14 | <TabAtkins> | jsbell: I find it makes it *extremely* clear that the sync part ends there. Definitely superfluous, but still helpful imo. |
| 00:15 | <Domenic> | jsbell: we've stopped using it as of the Infra Era, which formally defines return/throw to terminate the algorithm |
| 00:15 | <Domenic> | As with most things in Infra, HTML is gradually updating to use them as we refactor various algorithms, so it'll be a mismash for a while. |
| 00:16 | <Domenic> | https://infra.spec.whatwg.org/#algorithm-control-flow |
| 00:23 | <jsbell> | Thanks. "use judiciously when it improves readability" is a possible take-away, balancing the risk of making its absence confusing. |
| 00:30 | <tobie> | jsbell: it's fine until you have nested steps at which point it increases confusion |
| 04:24 | <MikeSmith> | annevk: the problems with instability of w3c-test.org are basically just that wptserve isn’t really suitable for being run as a persistent process. w3c-test.org is the only place where it’s run persistently with any kind fo load; everybody else just runs it locally with insignificant load |
| 04:25 | <MikeSmith> | annevk: the problem is that if wptserve running persistently under load leaks memory. Lots of it, til it eventually consumes 4GB+ of RAM. And it also starts to do other weird things |
| 04:26 | <MikeSmith> | annevk: so anyway I've now made a cron job that just restarts it every 1 hour |
| 06:07 | <jgraham> | As per the other channel, running a few thousand test files in CI doesn't seem like obviously insignificant load |
| 06:07 | <MikeSmith> | yeah |
| 06:07 | <jgraham> | So I wonder if there's something more specific that's problematic (e.g. specific tests that cause leaks) |
| 06:07 | <MikeSmith> | dunno |
| 06:07 | <MikeSmith> | very hard to troubleshoot |
| 06:10 | <jgraham> | I can maybe write a special patch to generate some memory logs |
| 06:11 | <jgraham> | See if we can figure out what's goign on |
| 06:12 | <MikeSmith> | that would be good |
| 06:14 | <MikeSmith> | when I restarted it today, on the w3c-test.org which has 7.8GB of RAM and normally has about 5.8GB of that sitting free, there was instead only 1.7GB free |
| 06:14 | <MikeSmith> | so hey on the bright side, it hadn’t yet consumed all of the free RAM |
| 07:21 | <annevk> | rbyers: I lost Chromium issue privileges so I can no longer add labels and such |
| 07:21 | <annevk> | rbyers: intentional? |
| 07:21 | <annevk> | rbyers: https://bugs.chromium.org/p/chromium/issues/detail?id=722737 will be the kind of issues you get from now |
| 08:20 | <jochen__> | sorry about that, I'll ping internally to restore your privs |
| 09:16 | <smaug> | annevk: https://html.spec.whatwg.org/#attr-label-for |
| 09:16 | <smaug> | "same tree" |
| 09:17 | <smaug> | but that doesn't care about in-document or anything |
| 09:23 | <smaug> | filed https://github.com/whatwg/html/issues/2686 |
| 09:29 | <Younder> | smaug, What is the reason for the current instability in the released blink implementation? |
| 09:29 | <smaug> | ? |
| 09:29 | smaug | doesn't know too much about blink |
| 09:30 | <smaug> | Last time I hacked webkit (pre-blink) was after all 11+ years ago :p |
| 10:55 | <annevk> | jochen__: do you think you could describe what document.open() does exactly in https://github.com/whatwg/html/issues/1698? |
| 10:55 | <annevk> | jochen__: I guess I can start writing some tests |
| 11:01 | <jochen__> | I can try, however, it's very different from what the spec says or Firefox does :/ |
| 11:05 | <rbyers> | annevk: Sorry, no of course it's not intentional 😉. I'll investigate. |
| 11:10 | <jochen__> | rbyers: i've sent an email to committers@ |
| 11:16 | <annevk> | jochen__: what Fx and the spec do is overly complicated so I'd rather not go there |
| 11:33 | <annevk> | nox: you said that WebKit doesn't implement document.open() with three arguments, but afaict it does |
| 11:33 | <annevk> | nox: in https://github.com/whatwg/html/issues/1698 |
| 11:33 | <annevk> | nox: at least document.open("","","") opens a new window in Safari TP |
| 11:40 | <nox> | annevk: Yeah that part was wrong |
| 11:41 | <annevk> | nox: would you mind if I remove the comment? |
| 11:41 | <nox> | annevk: Nope. |
| 11:42 | <annevk> | ta |
| 11:42 | <nox> | annevk: Is the spec going to align to WebKit behaviour? |
| 11:55 | <jochen__> | i hope not, unless WK meanwhile implements the origin check :) |
| 11:57 | <annevk> | nox: something that doesn't create a new global ideally |
| 12:02 | <nox> | annevk: WOOOO |
| 12:03 | <nox> | jochen__: What? |
| 12:09 | <annevk> | nox: it's probably gonna take a while longer, but I've made some progress |
| 12:38 | <jochen__> | nox, afaik WK hasn't implemented https://github.com/whatwg/html/pull/583 yet |
| 12:44 | <nox> | jochen__: Oh. |
| 12:44 | <nox> | jochen__: I meant the part where WK doesn't create a new Window. |
| 14:29 | <jochen__> | annevk: I tried to summarize what blink does... at places, it's not quite clear to me how our code translates to spec text and vice versa |
| 14:30 | <annevk> | jochen__: ta, I'll have a look tomorrow or maybe later today, time to do some parenting |
| 14:34 | <jochen__> | enjoy |
| 19:53 | <Yuhong> | It would be funny if W3C was to version HTML based on the WHATWG web developer edition. |
| 19:54 | <Yuhong> | Similar to what they did with HTML 3.2. |
| 21:33 | <zcorpan> | cdumez is quick with fixing bugs <3 https://bugs.webkit.org/show_bug.cgi?id=172114 |
| 21:43 | <zcorpan> | in related news https://bugs.webkit.org/show_bug.cgi?id=110001 - yay! |